The paper "PVNet: Pixel-wise Voting Network for 6DoF Pose Estimation" use RANSAC based voting for localizing keypoints, and further use PnP to calculate object poses.
The paper "6DoF Object Pose Estimation via Differentiable Proxy Voting Regularizer" employ Hough voting in testing to localize keypoints and then EPnP to solve 6DOF poses. It has been mentioned in "Implementation details" section of their paper. I'm not sure What do they employ in training to localize keypoints, whether Hough voting or RANSAC??
What's the difference b/w RANSAC-based voting & Hough voting to estimate vector-field for keypoints localization? Which one is better in which situations?